多时相SAR船只目标的特征点匹配方法对比分析

Comparative analysis of feature point matching methods for multi-temporal synthetic aperture radar ship target

多时相合成孔径雷达(SAR)卫星数据的组合能有效缩短重访周期、提高船只目标匹配跟踪能力,但多时相SAR工作参数(波段、分辨率等)、船只特点(尺寸小、常处于运动等)均会影响匹配跟踪方法的精度与效率。因此,将Harris方法、FAST(Features from Accelerated Segment Test)方法、尺度不变特征转换(SIFT)方法和SAR尺度不变特征转换(SAR-SIFT)这4种匹配方法应用至多时相SAR船只目标,从抗噪性、描述目标能力以及特征点稳定性三方面综合评估各方法的性能。在抗噪性方面,SAR-SIFT、FAST和Harris方法均未检测到海洋背景的干扰特征点;在描述目标能力方面,SAR-SIFT方法提取的特征点数比其他方法高出15%,并且特征点分布均匀;在特征点稳定性方面,SAR-SIFT方法的同名特征点重复检测率比其他方法高10%以上。实验结果表明,SAR-SIFT方法在多时相SAR船只目标特征点提取匹配中更具优势。

The combination of multi-temporal Synthetic Aperture Radar(SAR)satellite data can effectively shorten the revisit period and improve the ability of ship target matching and tracking.However,the accuracy and efficiency of the matching and tracking methods were affected by the multi-temporal SAR working parameters(band,resolution,etc.)and ship characteristics(small size,often in motion,etc.).Therefore,four matching methods,Harris,FAST(Features from Accelerated Segment Test),SIFT(Scale Invariant Feature Transform)and SAR-SIFT(Synthetic Aperture Radar-SIFT),were applied to multi-temporal SAR ship targets,and the performance of each algorithm was comprehensively evaluated from three aspects:anti-noise ability,target description ability and feature point stability.In terms of anti-noise ability,none noise feature points of background were detected by SAR-SIFT,FAST and Harris.In terms of target description ability,the number of feature points extracted by SAR-SIFT is 15%higher than those of other methods,and the feature points are evenly distributed.In terms of feature point stability,the repeat detection rate of the feature points for SAR-SIFT is more than 10%higher than that of other methods.The results show that the SAR-SIFT method is more advantageous in extracting and matching the target feature points of multi-temporal SAR ships.